An introduction to Generative AI

What is Generative AI?

Generative AI is a set of algorithms, capable of generating content—such as text, images, or audio.

For example, GPT-3.5, a model trained on large volumes of text, can answer questions, summarize text and perform sentiment analysis. DALL-E can transform text to images, expand images beyond their original size, and create variations.

What can Generative AI do?

Generate content and ideas: create new and unique outputs such as a video advertisement, or even design a new protein with antimicrobial properties.
Improve efficiency: accelerate manual or repetitive tasks such as writing emails, code, or summarize large documents.
Personalize experiences: Create content and information tailored to a specific audience, such as targeted advertisements based on patterns in a specific customer behaviour.

ChatGPT Capabilities

Use cases using RPA and Generative AI

Knowledge Management - Create your own knowledge base with full data control and confidentiality

Opportunity

Information is often spread around many documents: policies, manuals, reports, etc. To find what they are looking for, users would need to go through multiple documents. In addition, documents can be long, unstructured and in multiple formats.

 

We use a combination of RPA and GenAI to bring information together in a knowledge base, where users can ask questions in plain language. The model answers in a user-friendly way, and even understands follow-up questions. It also provides source references. An example is a Knowledge Bot we created for Child Welfare, where policies and regulations are updated frequently. Our solution continuously ingests the latest information, which users can easily access.

Related use case examples:

  1. Control and manage regulatory policies in a bank
  2. Review and compare legal contract versions
  3. Identify suitable candidates from a recruitment Database (e.g. specific skillset).
  4. Retrieve information across multiple (types of) documents; a specific use case for this would be a FAQ chatbot

Knowledge Diagramming - Distilling words into visualizations.

Opportunity

Business concepts can be complex and multifaceted. Sometimes these can be difficult to represent in just words.

Think about it, would you rather have a power-point presentation filled with words or diagrams?

Knowledge Diagrams can provide unified perspective of knowledge spread across dozens of different sources.

Related use case examples:

  1. Balance Scorecards from corporate and project documents
  2. Relationship maps of household participants from social worker case files
  3. Various business analysis diagrams used to analyse, plan, and track goals
  4. Process diagrams (see video below)

Customer Service Virtual Assistant - Customer Service Reps love this new way of working!

Opportunity

Responding to customer emails can be time consuming. Emails may come in high volumes, in different languages and often require time to read, understand and respond. Our solution can significantly enhance customer support operations by automating responses to common inquiries and providing timely assistance. The customer queries could include placing an order, change their address, make a complaint or asking for a refund. Our solution can read , understand and classifies the multilingual emails . It can then recommend an action and execute upon confirmation.

One such solution in Hospitality Industry is addressing customer emails by understanding the tone , sentiments , purpose and drafting a response for the customer service representative for their review . The response could be a ”Thank you” email, reply to the enquiry, reply with discount coupons and vouchers, addressing complaints and escalations.

Related use case examples:

  1. Client information or order updates such as a change of address
  2. Automated creation of service tickets for complaints
  3. Providing automated responses with service quotations and handling inquiries regarding payments
  4. Provide information and process bookings for hotels, restaurants, cruises, rental cars and other travel.

Invoice Processing Assistant - Easier to Train Cheaper to Implement

Opportunity

The challenge of efficiently managing invoice processing is a common one. Traditional methods often involve manual data entry, which can be time-consuming, error-prone, and costly. On the other hand, the popular Document Understanding approaches tend to require extended model training durations and high license costs, thus detracting from their cost-effectiveness.

Our innovative approach to enhancing Document Understanding processes integrates ChatGPT’s Natural Language Understanding and Conversation capability with UiPath’s enhanced Integration Service and powerful robots.

Related use case examples:

  1. Invoice Processing
  2. Proof of Delivery Confirmations
  3. Insurance Claims Processing
  4. Bank Statement Analysis

Test Automation: Test Data Generation - Automation Testing shouldn’t be done manually anymore!

Opportunity

Information is often spread around many documents: policies, manuals, reports, etc. To find what they are looking for, users would need to go through multiple documents. In addition, documents can be long, unstructured and in multiple formats.

We use a combination of RPA and GenAI to bring information together in a knowledge base, where users can ask questions in plain language. The model answers in a user-friendly way, and even understands follow-up questions. It also provides source references. An example is a Knowledge Bot we created for Child Welfare, where policies and regulations are updated frequently. Our solution continuously ingests the latest information, which users can easily access.

Related use case examples:

  1. SAP4 / HANA Migrations
  2. ERP implementations
  3. New software releases
  4. Test automation as part of video game development.